Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
#data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25
data_dir = './data'

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import helper
import os
from glob import glob
from matplotlib import pyplot
import numpy as np

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(np.array(helper.images_square_grid(mnist_images, 'L')))

<matplotlib.image.AxesImage at 0x7f5e7a1d6198>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
helper.images_square_grid(mnist_images, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7f5e7a176358>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

/usr/local/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: compiletime version 3.5 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.6
  return f(*args, **kwds)
/usr/local/lib/python3.6/site-packages/h5py/__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
  from ._conv import register_converters as _register_converters

TensorFlow Version: 1.4.1
Default GPU Device: /device:GPU:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    # TODO: Implement Function
    real_inputs = tf.placeholder(dtype=tf.float32, shape=(None, image_width, image_height, image_channels), name='real_input')
    z_inputs = tf.placeholder(dtype=tf.float32, shape=(None, z_dim), name='z_inputs')
    learning_rate = tf.placeholder(dtype=tf.float32, name='learning_rate')

    return real_inputs, z_inputs, learning_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False, alpha=0.2):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    # print("==discriminator structrue==")
    with tf.variable_scope('discriminator', reuse=reuse):        
        conv1 = tf.layers.conv2d(images, 32, 3, 2, 'SAME')
        lrelu1 = tf.maximum(alpha * conv1, conv1)
        # print(lrelu1.shape)
        
        conv2 = tf.layers.conv2d(lrelu1, 64, 3, 2, 'SAME')
        batch_norm2 = tf.layers.batch_normalization(conv2, training=True)
        lrelu2 = tf.maximum(alpha * batch_norm2, batch_norm2)
        # print(lrelu2.shape)
        
        conv3 = tf.layers.conv2d(lrelu2, 128, 3, 1, 'SAME')
        batch_norm3 = tf.layers.batch_normalization(conv3, training=True)
        lrelu3 = tf.maximum(alpha * batch_norm3, batch_norm3)
        # print(lrelu3.shape)
        
        features = tf.reduce_mean(lrelu3, (1, 2)) 
        # print(features.shape)
        # print()
        logits = tf.layers.dense(features, 1)
        
        out = tf.sigmoid(logits)
        
        return out, logits
        

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True, alpha=0.2):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    # print("==generator structrue==")
    with tf.variable_scope('generator', reuse=not is_train):
        x1 = tf.layers.dense(z, 7 * 7 * 512)
        
        x1 = tf.reshape(x1, (-1, 7, 7, 512))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        # print(x1.shape)
        
        x2 = tf.layers.conv2d_transpose(x1, 256, 3, strides=1, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(alpha * x2, x2)
        # print(x2.shape)
        
        x3 = tf.layers.conv2d_transpose(x2, 128, 3, strides=1, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        # print(x3.shape)
        
        x4 = tf.layers.conv2d_transpose(x3, 64, 3, strides=2, padding='same', kernel_initializer=tf.contrib.layers.xavier_initializer())
        x4 = tf.layers.batch_normalization(x4, training=is_train)
        x4 = tf.maximum(alpha * x4, x4)
        # print(x4.shape)
        
        # Output layer
        logits = tf.layers.conv2d_transpose(x4, out_channel_dim, 3, strides=2, padding='same')
        # print(logits.shape)
        # print()
        
        out = tf.tanh(logits)
        
        return out


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    g_model = generator(input_z, out_channel_dim)
    d_model_real, d_logits_real = discriminator(input_real)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True)
    
    d_loss_real = tf.reduce_mean(
                      tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, 
                                                              labels=tf.ones_like(d_logits_real) * 0.9))
    d_loss_fake = tf.reduce_mean(
                      tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, 
                                                              labels=tf.zeros_like(d_logits_real)))
    d_loss = d_loss_real + d_loss_fake

    g_loss = tf.reduce_mean(
                 tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake,
                                                         labels=tf.ones_like(d_logits_fake)))
    
    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]
    
    d_train_opt = tf.train.AdamOptimizer(
        learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
    
    ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
    g_updates = [opt for opt in ops if opt.name.startswith('generator')]
    with tf.control_dependencies(g_updates):
        g_train_opt = tf.train.AdamOptimizer(
            learning_rate, beta1).minimize(g_loss, var_list=g_vars)
        
    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    if cmap == 'gray':
        pyplot.imshow(np.array(helper.images_square_grid(samples, image_mode)))
        pyplot.show()
    else:
        images_grid = helper.images_square_grid(samples, image_mode)
        pyplot.imshow(images_grid, cmap=cmap)
        pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    # build model
    input_real, input_z, _ = model_inputs(data_shape[1], data_shape[2], data_shape[3], z_dim)
    d_loss, g_loss = model_loss(input_real, input_z, data_shape[3])
    d_opt, g_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    steps = 0
    losses = []
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            for batch_images in get_batches(batch_size):
                # TODO: Train Model
                batch_images *= 2
                steps += 1
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                
                _ = sess.run(d_opt, feed_dict={input_real: batch_images, input_z: batch_z})
                _ = sess.run(g_opt, feed_dict={input_z: batch_z})
                
                if steps % 10 == 0:
                    # At the end of every 10 epochs, get the losses and print them out
                    train_loss_d = d_loss.eval({input_z: batch_z, input_real: batch_images})
                    train_loss_g = g_loss.eval({input_z: batch_z})

                    print("Epoch {}/{}...".format(epoch_i+1, epochs),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))
                    losses.append((train_loss_d, train_loss_g))
                
                if steps % 100 == 0:
                    show_generator_output(sess, 25, input_z, data_shape[3], data_image_mode)
        return losses

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 128
z_dim = 128
learning_rate = 0.001
beta1 = 0.5


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 2

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    losses = train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode)

Epoch 1/2... Discriminator Loss: 1.2982... Generator Loss: 0.6907
Epoch 1/2... Discriminator Loss: 1.3219... Generator Loss: 0.8063
Epoch 1/2... Discriminator Loss: 1.3368... Generator Loss: 0.8252
Epoch 1/2... Discriminator Loss: 1.3449... Generator Loss: 0.8178
Epoch 1/2... Discriminator Loss: 1.3491... Generator Loss: 0.8076
Epoch 1/2... Discriminator Loss: 1.3522... Generator Loss: 0.7830
Epoch 1/2... Discriminator Loss: 1.3571... Generator Loss: 0.7813
Epoch 1/2... Discriminator Loss: 1.3564... Generator Loss: 0.7837
Epoch 1/2... Discriminator Loss: 1.3450... Generator Loss: 0.7864
Epoch 1/2... Discriminator Loss: 1.3386... Generator Loss: 0.8198

Epoch 1/2... Discriminator Loss: 1.3397... Generator Loss: 0.8069
Epoch 1/2... Discriminator Loss: 1.3467... Generator Loss: 0.7949
Epoch 1/2... Discriminator Loss: 1.3477... Generator Loss: 0.7947
Epoch 1/2... Discriminator Loss: 1.3542... Generator Loss: 0.7675
Epoch 1/2... Discriminator Loss: 1.3389... Generator Loss: 0.8057
Epoch 1/2... Discriminator Loss: 1.3401... Generator Loss: 0.7847
Epoch 1/2... Discriminator Loss: 1.3409... Generator Loss: 0.7763
Epoch 1/2... Discriminator Loss: 1.3423... Generator Loss: 0.7930
Epoch 1/2... Discriminator Loss: 1.3435... Generator Loss: 0.7691
Epoch 1/2... Discriminator Loss: 1.3512... Generator Loss: 0.7950

Epoch 1/2... Discriminator Loss: 1.3450... Generator Loss: 0.8017
Epoch 1/2... Discriminator Loss: 1.3584... Generator Loss: 0.7863
Epoch 1/2... Discriminator Loss: 1.3505... Generator Loss: 0.7747
Epoch 1/2... Discriminator Loss: 1.3314... Generator Loss: 0.8296
Epoch 1/2... Discriminator Loss: 1.3495... Generator Loss: 0.7626
Epoch 1/2... Discriminator Loss: 1.3377... Generator Loss: 0.8015
Epoch 1/2... Discriminator Loss: 1.3499... Generator Loss: 0.8207
Epoch 1/2... Discriminator Loss: 1.3513... Generator Loss: 0.7950
Epoch 1/2... Discriminator Loss: 1.3429... Generator Loss: 0.8058
Epoch 1/2... Discriminator Loss: 1.3332... Generator Loss: 0.8074

Epoch 1/2... Discriminator Loss: 1.3386... Generator Loss: 0.7856
Epoch 1/2... Discriminator Loss: 1.3339... Generator Loss: 0.8027
Epoch 1/2... Discriminator Loss: 1.3187... Generator Loss: 0.7937
Epoch 1/2... Discriminator Loss: 1.3304... Generator Loss: 0.8209
Epoch 1/2... Discriminator Loss: 1.3227... Generator Loss: 0.7902
Epoch 1/2... Discriminator Loss: 1.3327... Generator Loss: 0.8145
Epoch 1/2... Discriminator Loss: 1.3247... Generator Loss: 0.8132
Epoch 1/2... Discriminator Loss: 1.3252... Generator Loss: 0.7961
Epoch 1/2... Discriminator Loss: 1.3189... Generator Loss: 0.7886
Epoch 1/2... Discriminator Loss: 1.3364... Generator Loss: 0.8878

Epoch 1/2... Discriminator Loss: 1.3103... Generator Loss: 0.8339
Epoch 1/2... Discriminator Loss: 1.2941... Generator Loss: 0.8507
Epoch 1/2... Discriminator Loss: 1.2956... Generator Loss: 0.8513
Epoch 1/2... Discriminator Loss: 1.3234... Generator Loss: 0.8296
Epoch 1/2... Discriminator Loss: 1.3107... Generator Loss: 0.8291
Epoch 1/2... Discriminator Loss: 1.2871... Generator Loss: 0.8723
Epoch 2/2... Discriminator Loss: 1.3156... Generator Loss: 0.8084
Epoch 2/2... Discriminator Loss: 1.3224... Generator Loss: 0.7881
Epoch 2/2... Discriminator Loss: 1.2865... Generator Loss: 0.7982
Epoch 2/2... Discriminator Loss: 1.2778... Generator Loss: 0.8410

Epoch 2/2... Discriminator Loss: 1.2724... Generator Loss: 0.8573
Epoch 2/2... Discriminator Loss: 1.2673... Generator Loss: 0.8540
Epoch 2/2... Discriminator Loss: 1.2924... Generator Loss: 0.8949
Epoch 2/2... Discriminator Loss: 1.2435... Generator Loss: 0.8957
Epoch 2/2... Discriminator Loss: 1.2939... Generator Loss: 0.7934
Epoch 2/2... Discriminator Loss: 1.2884... Generator Loss: 0.7933
Epoch 2/2... Discriminator Loss: 1.2448... Generator Loss: 0.8560
Epoch 2/2... Discriminator Loss: 1.2662... Generator Loss: 0.8161
Epoch 2/2... Discriminator Loss: 1.2405... Generator Loss: 0.9080
Epoch 2/2... Discriminator Loss: 1.2298... Generator Loss: 0.9040

Epoch 2/2... Discriminator Loss: 1.2432... Generator Loss: 0.9029
Epoch 2/2... Discriminator Loss: 1.2457... Generator Loss: 0.9056
Epoch 2/2... Discriminator Loss: 1.2458... Generator Loss: 0.8450
Epoch 2/2... Discriminator Loss: 1.2370... Generator Loss: 0.8878
Epoch 2/2... Discriminator Loss: 1.2339... Generator Loss: 0.7866
Epoch 2/2... Discriminator Loss: 1.2002... Generator Loss: 0.9493
Epoch 2/2... Discriminator Loss: 1.2327... Generator Loss: 0.8615
Epoch 2/2... Discriminator Loss: 1.2370... Generator Loss: 0.9053
Epoch 2/2... Discriminator Loss: 1.2069... Generator Loss: 0.8275
Epoch 2/2... Discriminator Loss: 1.1975... Generator Loss: 0.8665

Epoch 2/2... Discriminator Loss: 1.2084... Generator Loss: 0.8695
Epoch 2/2... Discriminator Loss: 1.2274... Generator Loss: 0.8328
Epoch 2/2... Discriminator Loss: 1.2861... Generator Loss: 0.8626
Epoch 2/2... Discriminator Loss: 1.1867... Generator Loss: 0.9138
Epoch 2/2... Discriminator Loss: 1.1624... Generator Loss: 0.9345
Epoch 2/2... Discriminator Loss: 1.2120... Generator Loss: 0.8234
Epoch 2/2... Discriminator Loss: 1.2000... Generator Loss: 0.9380
Epoch 2/2... Discriminator Loss: 1.1684... Generator Loss: 0.8673
Epoch 2/2... Discriminator Loss: 1.2067... Generator Loss: 0.8011
Epoch 2/2... Discriminator Loss: 1.1910... Generator Loss: 0.9255

Epoch 2/2... Discriminator Loss: 1.2131... Generator Loss: 0.7726
Epoch 2/2... Discriminator Loss: 1.2638... Generator Loss: 0.7931
Epoch 2/2... Discriminator Loss: 1.1763... Generator Loss: 0.8145
Epoch 2/2... Discriminator Loss: 1.2196... Generator Loss: 0.9508
Epoch 2/2... Discriminator Loss: 1.2480... Generator Loss: 0.9262
Epoch 2/2... Discriminator Loss: 1.1641... Generator Loss: 0.8571
Epoch 2/2... Discriminator Loss: 1.1679... Generator Loss: 0.8989
Epoch 2/2... Discriminator Loss: 1.1507... Generator Loss: 1.0425
Epoch 2/2... Discriminator Loss: 1.2142... Generator Loss: 0.8419
Epoch 2/2... Discriminator Loss: 1.1532... Generator Loss: 0.9188

Epoch 2/2... Discriminator Loss: 1.2351... Generator Loss: 0.9089
Epoch 2/2... Discriminator Loss: 1.1948... Generator Loss: 0.8859
Epoch 2/2... Discriminator Loss: 1.0539... Generator Loss: 1.0271

from matplotlib import pyplot as plt
fig, ax = plt.subplots()
losses = np.array(losses)
plt.plot(losses.T[0], label='Discriminator', alpha=0.5)
plt.plot(losses.T[1], label='Generator', alpha=0.5)
plt.title("Training Losses")
plt.legend()

<matplotlib.legend.Legend at 0x7f5d7c3fcda0>

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(np.array(helper.images_square_grid(mnist_images, 'L')))

<matplotlib.image.AxesImage at 0x7f5d7c3f2278>

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 64
z_dim = 256
learning_rate = 0.001
beta1 = 0.4


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 10

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    losses = train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches, celeba_dataset.shape, celeba_dataset.image_mode)

Epoch 1/10... Discriminator Loss: 1.3181... Generator Loss: 0.7756
Epoch 1/10... Discriminator Loss: 1.2662... Generator Loss: 0.8238
Epoch 1/10... Discriminator Loss: 1.3612... Generator Loss: 0.7925
Epoch 1/10... Discriminator Loss: 1.3688... Generator Loss: 0.8120
Epoch 1/10... Discriminator Loss: 1.3978... Generator Loss: 0.7938
Epoch 1/10... Discriminator Loss: 1.3824... Generator Loss: 0.7857
Epoch 1/10... Discriminator Loss: 1.3900... Generator Loss: 0.7699
Epoch 1/10... Discriminator Loss: 1.3851... Generator Loss: 0.7877
Epoch 1/10... Discriminator Loss: 1.3667... Generator Loss: 0.7999
Epoch 1/10... Discriminator Loss: 1.3696... Generator Loss: 0.7737

Epoch 1/10... Discriminator Loss: 1.3679... Generator Loss: 0.7994
Epoch 1/10... Discriminator Loss: 1.3865... Generator Loss: 0.7775
Epoch 1/10... Discriminator Loss: 1.3601... Generator Loss: 0.7991
Epoch 1/10... Discriminator Loss: 1.3620... Generator Loss: 0.7916
Epoch 1/10... Discriminator Loss: 1.3717... Generator Loss: 0.7843
Epoch 1/10... Discriminator Loss: 1.3822... Generator Loss: 0.7796
Epoch 1/10... Discriminator Loss: 1.3773... Generator Loss: 0.7962
Epoch 1/10... Discriminator Loss: 1.3628... Generator Loss: 0.7736
Epoch 1/10... Discriminator Loss: 1.3736... Generator Loss: 0.8042
Epoch 1/10... Discriminator Loss: 1.3706... Generator Loss: 0.7826

Epoch 1/10... Discriminator Loss: 1.4032... Generator Loss: 0.7581
Epoch 1/10... Discriminator Loss: 1.3911... Generator Loss: 0.7711
Epoch 1/10... Discriminator Loss: 1.3859... Generator Loss: 0.8022
Epoch 1/10... Discriminator Loss: 1.3759... Generator Loss: 0.7742
Epoch 1/10... Discriminator Loss: 1.3461... Generator Loss: 0.8059
Epoch 1/10... Discriminator Loss: 1.3890... Generator Loss: 0.7730
Epoch 1/10... Discriminator Loss: 1.3376... Generator Loss: 0.7976
Epoch 1/10... Discriminator Loss: 1.3686... Generator Loss: 0.7686
Epoch 1/10... Discriminator Loss: 1.4175... Generator Loss: 0.7454
Epoch 1/10... Discriminator Loss: 1.3748... Generator Loss: 0.7649

Epoch 1/10... Discriminator Loss: 1.3292... Generator Loss: 0.7783
Epoch 1/10... Discriminator Loss: 1.3837... Generator Loss: 0.7649
Epoch 1/10... Discriminator Loss: 1.3556... Generator Loss: 0.7932
Epoch 1/10... Discriminator Loss: 1.3735... Generator Loss: 0.7836
Epoch 1/10... Discriminator Loss: 1.3598... Generator Loss: 0.8221
Epoch 1/10... Discriminator Loss: 1.3366... Generator Loss: 0.8038
Epoch 1/10... Discriminator Loss: 1.3554... Generator Loss: 0.7916
Epoch 1/10... Discriminator Loss: 1.3605... Generator Loss: 0.8116
Epoch 1/10... Discriminator Loss: 1.3668... Generator Loss: 0.8058
Epoch 1/10... Discriminator Loss: 1.3948... Generator Loss: 0.7755

Epoch 1/10... Discriminator Loss: 1.3450... Generator Loss: 0.7930
Epoch 1/10... Discriminator Loss: 1.3657... Generator Loss: 0.7838
Epoch 1/10... Discriminator Loss: 1.3942... Generator Loss: 0.7661
Epoch 1/10... Discriminator Loss: 1.3458... Generator Loss: 0.7821
Epoch 1/10... Discriminator Loss: 1.3885... Generator Loss: 0.7847
Epoch 1/10... Discriminator Loss: 1.3980... Generator Loss: 0.7859
Epoch 1/10... Discriminator Loss: 1.3745... Generator Loss: 0.7705
Epoch 1/10... Discriminator Loss: 1.3482... Generator Loss: 0.8087
Epoch 1/10... Discriminator Loss: 1.3756... Generator Loss: 0.7694
Epoch 1/10... Discriminator Loss: 1.3474... Generator Loss: 0.7672

Epoch 1/10... Discriminator Loss: 1.3770... Generator Loss: 0.7846
Epoch 1/10... Discriminator Loss: 1.3678... Generator Loss: 0.7754
Epoch 1/10... Discriminator Loss: 1.3814... Generator Loss: 0.7920
Epoch 1/10... Discriminator Loss: 1.3924... Generator Loss: 0.7710
Epoch 1/10... Discriminator Loss: 1.3800... Generator Loss: 0.7881
Epoch 1/10... Discriminator Loss: 1.3749... Generator Loss: 0.7787
Epoch 1/10... Discriminator Loss: 1.3881... Generator Loss: 0.7916
Epoch 1/10... Discriminator Loss: 1.3833... Generator Loss: 0.7648
Epoch 1/10... Discriminator Loss: 1.3936... Generator Loss: 0.7729
Epoch 1/10... Discriminator Loss: 1.3625... Generator Loss: 0.7868

Epoch 1/10... Discriminator Loss: 1.3680... Generator Loss: 0.7730
Epoch 1/10... Discriminator Loss: 1.3387... Generator Loss: 0.8079
Epoch 1/10... Discriminator Loss: 1.3505... Generator Loss: 0.8178
Epoch 1/10... Discriminator Loss: 1.3637... Generator Loss: 0.7513
Epoch 1/10... Discriminator Loss: 1.3309... Generator Loss: 0.8098
Epoch 1/10... Discriminator Loss: 1.3738... Generator Loss: 0.7745
Epoch 1/10... Discriminator Loss: 1.4134... Generator Loss: 0.7299
Epoch 1/10... Discriminator Loss: 1.3757... Generator Loss: 0.7739
Epoch 1/10... Discriminator Loss: 1.3891... Generator Loss: 0.7610
Epoch 1/10... Discriminator Loss: 1.4395... Generator Loss: 0.7445

Epoch 1/10... Discriminator Loss: 1.3445... Generator Loss: 0.7848
Epoch 1/10... Discriminator Loss: 1.3647... Generator Loss: 0.7542
Epoch 1/10... Discriminator Loss: 1.4178... Generator Loss: 0.7633
Epoch 1/10... Discriminator Loss: 1.3747... Generator Loss: 0.7520
Epoch 1/10... Discriminator Loss: 1.3842... Generator Loss: 0.7828
Epoch 1/10... Discriminator Loss: 1.3845... Generator Loss: 0.7803
Epoch 1/10... Discriminator Loss: 1.3665... Generator Loss: 0.7931
Epoch 1/10... Discriminator Loss: 1.4039... Generator Loss: 0.7647
Epoch 1/10... Discriminator Loss: 1.3887... Generator Loss: 0.7751
Epoch 1/10... Discriminator Loss: 1.3897... Generator Loss: 0.7636

Epoch 1/10... Discriminator Loss: 1.3617... Generator Loss: 0.7896
Epoch 1/10... Discriminator Loss: 1.4217... Generator Loss: 0.7441
Epoch 1/10... Discriminator Loss: 1.3795... Generator Loss: 0.7775
Epoch 1/10... Discriminator Loss: 1.3865... Generator Loss: 0.7718
Epoch 1/10... Discriminator Loss: 1.3788... Generator Loss: 0.7754
Epoch 1/10... Discriminator Loss: 1.3808... Generator Loss: 0.7848
Epoch 1/10... Discriminator Loss: 1.3909... Generator Loss: 0.7850
Epoch 1/10... Discriminator Loss: 1.3792... Generator Loss: 0.7699
Epoch 1/10... Discriminator Loss: 1.3983... Generator Loss: 0.7701
Epoch 1/10... Discriminator Loss: 1.3889... Generator Loss: 0.7733

Epoch 1/10... Discriminator Loss: 1.4192... Generator Loss: 0.7621
Epoch 1/10... Discriminator Loss: 1.3811... Generator Loss: 0.7828
Epoch 1/10... Discriminator Loss: 1.4007... Generator Loss: 0.7782
Epoch 1/10... Discriminator Loss: 1.3933... Generator Loss: 0.7747
Epoch 1/10... Discriminator Loss: 1.3739... Generator Loss: 0.7804
Epoch 1/10... Discriminator Loss: 1.3831... Generator Loss: 0.7867
Epoch 1/10... Discriminator Loss: 1.3736... Generator Loss: 0.7791
Epoch 1/10... Discriminator Loss: 1.3798... Generator Loss: 0.7893
Epoch 1/10... Discriminator Loss: 1.3822... Generator Loss: 0.7913
Epoch 1/10... Discriminator Loss: 1.3844... Generator Loss: 0.7811

fig, ax = plt.subplots()
losses = np.array(losses)
plt.plot(losses.T[0], label='Discriminator', alpha=0.5)
plt.plot(losses.T[1], label='Generator', alpha=0.5)
plt.title("Training Losses")
plt.legend()

<matplotlib.legend.Legend at 0x7f5cdbff5828>

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
helper.images_square_grid(mnist_images, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7f5cdbfd74a8>

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.